# Query and analyze your data using AI tools and Databox MCP The Databox Model Context Protocol (MCP) server lets you connect AI assistants like Claude, ChatGPT, and Gemini directly to your Databox account, so you can interact with your business data using natural language without leaving your AI chat interface. MCP standardizes how AI systems discover available tools, understand their capabilities, and execute actions on your behalf, making it possible to ask questions, fetch data, and trigger workflows in Databox while keeping access controlled and auditable. ## MCP overview Model Context Protocol (MCP) is an open standard that allows AI assistants to securely connect to external data sources and tools. Think of it as a bridge that lets your AI assistant access and interact with your Databox account while you chat. With Databox MCP, you can: - **Query your data** using natural language instead of building metrics manually - **Get instant insights** by asking questions like "What were our top traffic sources last month?" - **Automate data tasks** such as creating [data sources](/add-a-data-source), [ingesting data](/send-data-to-databox-via-api), or other supported actions - **Contextualize Databox data** using other tools connected to your AI assistant ## MCP tools overview MCP tools are the individual actions exposed by an MCP server. Each tool has a clear name, description, and input schema so an AI model knows when and how to use it. In Databox MCP, tools are used when a request requires live data or an action that cannot be answered from the model's existing context. For example: - Retrieving metric values or dataset rows - Listing available metrics or [datasets](/understanding-datasets) - Validating inputs before running an operation You can view the complete and up-to-date list of available Databox MCP tools in the official [developer documentation](https://developers.databox.com/docs/mcp/tools). MCP tools do not give models unrestricted access. Each tool enforces Databox permissions and only exposes what the authenticated user is allowed to see or do. The model decides when to call a tool based on the user's request. If the request can be answered without Databox data, no tool is used. If Databox data is required, the model selects the appropriate MCP tool and sends a structured request. For example: - When you ask **"What columns are in my sales dataset?"** → The assistant uses the `ask_genie` tool - When you request **"Show me Google Analytics sessions for January"** → The assistant uses `list_metrics` and `load_metric_data` tools - When you say **"Create a new data source for marketing data"** → The assistant uses the `create_data_source` tool ## Connect an AI or automation tool to Databox MCP You can connect multiple AI models and automation platforms to the Databox MCP server. The setup flow is similar across clients and is based on secure authentication and MCP configuration. Some AI and automation tools restrict MCP connector access to specific plans, user roles, or workspace settings. In some cases, a workspace administrator must enable required features before other users can connect Databox MCP. For details on tool-specific requirements, refer to the Databox [developer documentation](https://developers.databox.com/docs/mcp/setup). ### ChatGPT 1. Open Settings in ChatGPT by clicking your profile menu and selecting **Settings**. 2. Go to **Apps** to manage integrations. 3. Open **Advanced settings** to view additional configuration options. 4. Enable **Developer Mode (Beta)** to allow custom MCP connectors and tools. 5. Click **Create app** to add a new connector for Databox MCP. 6. Enter the app details in the **New App** setup screen: - **Name:** Databox - **MCP server URL:** `https://mcp.databox.com/mcp` - **Authorization:** OAuth 2.0 7. Authenticate with Databox when prompted to allow ChatGPT to securely call the Databox MCP on your behalf. Connect Databox MCP to ChatGPT ### Claude 1. Open Settings in Claude by clicking your profile menu and selecting **Settings**. 2. Go to **Connectors** to manage integrations. 3. Click **Add custom connector** to create a new connector for Databox MCP. 4. Enter the connector details in the **Add custom connector** window: - **Name:** Databox - **Remote MCP server URL:** `https://mcp.databox.com/mcp` 5. Authenticate with Databox when prompted to allow Claude to securely call the Databox MCP on your behalf. Connect Databox MCP to Claude ### Gemini Gemini does not currently support MCP directly. To use Gemini models with Databox MCP, you need to run [Goose](https://block.github.io/goose/docs/category/getting-started/) locally as an MCP client. 1. Create a Gemini API key by clicking **Create API key** in the Gemini console. 2. Name the API key by entering a descriptive name in the text field. 3. Create a new project for the API key by clicking **Create project**. 4. Enter a project name and click **Create project** to finalize it. 5. Copy the API key by clicking the **copy icon**. 6. Open the Goose app and **paste your Gemini API key**, then confirm. 7. Select a Gemini model and click **Select model** to continue. 8. Open **Extensions** in Goose and click **Add custom extension**. 9. Enter the extension details: - **Extension name:** Databox - **Type:** HTTP - **Endpoint:** `https://mcp.databox.com/mcp` 10. Click **Add Extension** to register Databox MCP with Goose and Gemini. 11. Start a new chat and click the **dumbbell icon**. 12. **Enable Databox**, which opens a browser window for authentication. 13. **Log in to Databox** to authorize the connection. Connect Databox MCP to Gemini ### n8n 1. Create a new workflow and click the **plus icon** to add the first step. 2. Search for AI agent and select **AI Agent** to add it to the workflow. 3. Return to the canvas to open the workflow editor. 4. Click the **plus icon** on the AI Agent node to add a tool. 5. Select **MCP Client Tool** to connect the AI agent to Databox MCP. 6. Enter the Databox MCP server URL in the **Endpoint** field: `https://mcp.databox.com/mcp`. 7. Open the **Authentication** dropdown and select **MCP OAuth2**. 8. Select a credential for MCP OAuth2 API, or click **Create new credential** to add one. 9. Paste the Databox MCP server URL again in the **Server URL** field. 10. Click **Connect my account** to start the OAuth sign-in flow. 11. Sign in to Databox and authorize the connection in the pop-up window, then close it. 12. Click **Execute step** to establish the connection. 13. Open the **Tool name** dropdown and select an MCP tool operation. 14. Select `list_accounts` and execute the step to verify the connection. 15. Return to the canvas to continue building your workflow. Connect Databox MCP to Gemini Yes. The MCP server includes tools for creating, updating, and deleting data sources and datasets. Destructive actions are only performed when you explicitly request them, such as “delete this dataset.” Always review and confirm AI-suggested actions before proceeding. No. Once the connection is set up, you can interact with Databox using natural language through your AI assistant. Coding is not required for basic usage. No. MCP tools are built on top of the Databox APIs. They provide a standardized way for AI systems to use Databox without requiring custom integrations. Claude (Anthropic) and ChatGPT (OpenAI) are SOC 2 certified, meaning they meet established standards for data security, availability, and confidentiality. When using Databox MCP, these AI tools do not store your Databox data. They retrieve data on demand through secure, authenticated requests. All access is governed by Databox's existing permission model, so the AI can only access data that the authenticated user is already authorized to view. For more details on privacy and security considerations, refer to the [Databox MCP Security documentation](https://developers.databox.com/docs/mcp/security). There is no additional usage cost on the Databox side for using the MCP server beyond your existing Databox subscription. However, your AI tool or automation platform may have its own subscription fees, usage limits, or API costs that apply when making MCP requests. Databox native integrations automatically pull data from platforms like Google Analytics or Salesforce into Databox. The Databox MCP server lets AI assistants interact with your Databox account by querying data, creating resources, and analyzing metrics using natural language, acting as a conversational interface to your Databox data.   Ask Genie Get instant answers or help with your data using the in-app AI assistant. Talk to an expert For customers: Get help with your setup, strategy, or making the most of Databox. Book a demo New to Databox? See how it works and get guidance on getting started. Send an email Reach out to support for help with your account, data, or technical issues.